13
$\begingroup$

I tried to create a band structure with the mBJ-LDA method using VASP. I found information that the KPOINTS file should be created according to the "0-weight (Fake) SC procedure" (as for HSE).

I have a problem because when I approach the $\mathbf{k}$-points corresponding to the second block, I get a structure in which only the points from the second block are contained, not the path between them. How can I increase the accuracy of the mapping of the band structure? Thank you for your help, it's very important to me.

enter image description here

$\endgroup$
1
  • $\begingroup$ +1. Thanks for asking here, and we hope to see much more of you! Thank you for taking ProfM's advice and asking this as a separate question rather than trying to ask it in a comment to someone else's question. $\endgroup$ Aug 15, 2020 at 18:42

2 Answers 2

13
$\begingroup$

Here I will take Mg3Sb2 from Materials Project as an example to demonstrate the bandstructure calculation flow with the MBJ method.

  • (I) structure relaxation

  • (II) PBE band calculation based on the relaxed structure

    • (a) self-consistent (SCF) calculation [10_scf_cal].
    • (b) band calculation based on converged charge density at (II-a) [11_band_cal].
    • (c) read and plot EIGENVALUE.

enter image description here

  • (III) MBJ band calculation based on step (II).

    • (a) PBE SCF calculation with zero weight k-points.

      HOW TO WRITE KPOINTS

       (1) Copy the IBZKPT at step (II-a) into KPOINTS
       (2) Find all k-points from OUTCAR at step (II-b) and add to the KPOINTS with zero weight.
      

      INCAR

       ISTART=0
       ICHARG=2
       ENCUT=400
       EDIFF=1E-6
       NELM=300
       ISMEAR=-5
       IALGO=38
      
    • (b) MBJ SCF calculation with zero weight k-points.

      INCAR

       ### Read PBE wavefunction
       ISTART=1
       ICHARG=0
       ENCUT=400
       EDIFF=1E-6
       NELM=300
       ISMEAR=-5
      
       ###MBJ:
       METAGGA=MBJ
       LASPH=.TRUE.
       LMAXMIX=4
       GGA=CA
       #CMBJ=XXX     
      
    • (c) read and plot the EIGENVALE with zero weight k-point. enter image description here

  • PS: all related input files can be accessed through this link: https://github.com/Kohn-X/MBJ_cal

$\endgroup$
7
  • $\begingroup$ could you also please provide us with the INCAR settings during this procedure ? $\endgroup$
    – Chi Kou
    Feb 2, 2021 at 14:36
  • 1
    $\begingroup$ @ChiKou Later I will add the INCAR file. $\endgroup$
    – Jack
    Feb 2, 2021 at 22:44
  • 1
    $\begingroup$ @ChiKou All related input cards have been attached. $\endgroup$
    – Jack
    Feb 4, 2021 at 3:19
  • 1
    $\begingroup$ @ChiKou I have detailed my answer again. $\endgroup$
    – Jack
    Feb 16, 2021 at 3:40
  • 1
    $\begingroup$ @ChiKou All input files have been attached: github.com/Kohn-X/MBJ_cal $\endgroup$
    – Jack
    Feb 18, 2021 at 4:46
9
$\begingroup$

If you give the explicit points in the KPOINTS file in VASP for a band structure calculation, for example as required for hybrid functionals, the bands will only be calculated at the explicit $\mathbf{k}$-points you list. For example, if you have a cubic cell and want the path between $\Gamma$ at $(0,0,0)$ and X at $(0.5,0,0)$, then simply writing:

$$ 0.0 \,\, 0.0 \,\, 0.0 \\ 0.5 \,\, 0.0 \,\, 0.0 $$

in your KPOINTS file will calculate the bands at these two points, but not at the $\mathbf{k}$-points in between. To get more $\mathbf{k}$-points between these two points, you need to explicitly include them in your list. For example, to include five points when going from $\Gamma$ to X in the cubic cell you should write:

$$ 0.0 \,\, 0.0 \,\, 0.0 \\ 0.1 \,\, 0.0 \,\, 0.0 \\ 0.2 \,\, 0.0 \,\, 0.0 \\ 0.3 \,\, 0.0 \,\, 0.0 \\ 0.4 \,\, 0.0 \,\, 0.0 \\ 0.5 \,\, 0.0 \,\, 0.0 $$

This will calculate the bands at each of these points. For finer $\mathbf{k}$-point sampling, you simply need to increase the number of divisions between the path end-points.

If you are not using a hybrid functional, then you can use the "Line-mode" in the KPOINTS file where you only need to specify the end points of the path and VASP automatically calculates the bands at the extra points between the end points.

$\endgroup$
2
  • $\begingroup$ I think this is not an easy job to prepare the KPOINTS file for the hybrid calculations. I still doubt how can we prepare this file for some different K-points, like (0.0 0 0) to 0.3333, 0.5 0.5). Can we use vaspkit for it? some people use these points from OUTCAR of fake-scf calculations. Please generalize your answer so that we can learn it more. $\endgroup$
    – astha
    Sep 22, 2021 at 6:10
  • $\begingroup$ I do not understand why you mentioned GGA=CA in III(b) while you have provided POTCAR for PBE. Can you please explain it a bit more? $\endgroup$
    – astha
    Apr 9, 2022 at 2:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .